جستجوی مقالات مرتبط با کلیدواژه « مدل های علیتی » در نشریات گروه « پزشکی »

Generalized Structural Equation Modeling (GSEM) is a family of statistical techniques utilized in the analysis of multivariate, categorical and ordinal data in order to measure latent variables and their connection with each other. The aim of this study is to consider the structure of data, and introducing GSEM to medical science researchers and presenting a practical example of in medical science researches.
An introduction to Structural Equation Modeling (SEM), along with its advantages and disadvantages was presented, and also GSEM and its all kind of forms was specified. An example to study hypertension risk factors in patients suffering from diabetes was carried out, which was a demonstration of using GSEM method for binary response variables. The data includes a random sample of 2716 people from Golestan province cohort studies.
Age, body mass index, abdominal obesity, residence place, socioeconomic status, salt intake had direct effect on hypertension. Race, education, vitamin D and physical activity had direct and reverse effect on hypertension (p.value Unlike SEM, the limitative hypothesis that our data should have a normal distribution do not needed in this model, also GSEM is powerful tool in the analysis of categorized data. Nevertheless this method cannot perform goodness of fit test, and adjustment and modification method of the model directly, and that they are some limitation in using this method.

To develop a practical guide for health professionals who want to design appropriate interventions in health programs by means of organized research? Specific objectives included finding the most practical and accurate methods for decomposition of outcomes, measuring the impact of each determinant in applied settings, and prioritizing research for achieving this goal.

In this paper all causal models were reviewed (i.e. graphical, structural, sufficient-component cause and counterfactual models), and judged as to whether they can provide an appropriate conceptual framework to prioritize research and interventions.

The Attributable Fraction is the classic epidemiologic measure for causal attribution of a risk factor. Since Attributable Fraction depends on measuring zero levels, which is neither feasible nor desirable, "Generalized Impact Fraction" was substituted for this purpose. The appropriate conceptual framework in practical settings could be Counterfactual Analysis for comparative risk assessment.

As it is not possible to perform every theoretical approach, choosing the most practical methods is essential in selecting epiderniologic methods for health programs. It seems that despite its limitations, Counterfactual Analysis and subsequent "General Impact Fraction" estimation could be most practical and accurate way in this situation. Further development of the method can be accompanied by selection of a sequence of different interventions based on "Average Attributable Fraction", provided the required data are available.